OLD TYPES AND NEW 199 



the very peculiar device of dermal spines. There is no 

 reason, however, for regarding this character as due 

 to descent from a common spiny ancestor. It is not 

 reversion to an ancestral type, but rather a case of ^ 

 convergent variation. Similarity does not always indi- 

 cate genetic continuity. 



In the case of birds albinism, melanism and flavism 

 are modifications of ordinary pigmentation which 

 appear irregularly among many different species as 

 pathological "sports," but no one of these conditions 

 can be regarded as reversions to ancestral white, black, 

 or yellow types. 



E. REGRESSION 



Galton's "law of regression" refers to the widespread 

 phenomenon already explained of a constant swinging 

 back to mediocrity which the breeder must oppose with 

 continual selection in order to maintain the standard 

 of any particular strain. We have seen that within 

 a "pure line," regression is complete and that in popu- 

 lations made up of a mixture of pure lines it is a fac- 

 tor that must invariably be considered. Regression, 

 however, has to do with fluctuating variations and does 

 not bring about a permanent change of type. It 

 should, therefore, not be confused with reversion. 



3. EXPLANATION OF REVERSION 



Darwin, who did not always differentiate between 

 reversion and atavism, suggested that reversion was 

 due sometimes to the action of a more natural en- 



